AI Data Center Boom
As turbine delivery times stretch to six years, developers such as Crusoe are turning to aeroderivative engines, refurbished equipment, and alternative power systems to keep AI projects moving.
Crusoe’s April order for 13 aeroderivative gas turbines from ProEnergy captures a shift in the economics and logistics of AI infrastructure. Based on GE Aerospace jet engines, the units have a combined generation capacity of 650 megawatts and are intended for Crusoe’s US AI data center projects. The order does not mean the turbines are already installed or generating electricity.
As companies invest in chips, servers, and networking equipment, access to power generation has become another constraint on construction. The AI data center power challenge is no longer simply securing computing hardware. Developers also need generation equipment, grid connections, permits, and the operational systems that make large facilities dependable.
According to Wood Mackenzie analyst Aurora Tenorio, delivery times for gas turbines in the United States now average roughly six years, with some discussions extending into 2030 and 2031.
Are you looking for developers?
Global turbine orders exceed 100 gigawatts, while estimated annual manufacturing capacity stands at around 60 to 70 gigawatts. Capacity could reach approximately 87 gigawatts by 2028, but projected demand could rise to 129 gigawatts. These figures describe orders, production capacity, and forecasts, not equipment already supplying electricity.
AI data centers are adding to demand. Jefferies estimates that US data centers account for nearly 17 gigawatts of turbine demand in 2026, potentially more than tripling by 2028.
The pressure extends beyond technology developers. Utilities and rural electric cooperatives are also contending with longer lead times and higher equipment prices. An expanding data center pipeline competes for industrial components needed elsewhere in the power system.
The wait for new turbines is pushing developers toward used, refurbished, and repurposed units. The secondary market is active, but the number of available machines is limited. Analysts caution that it is unlikely to offset a meaningful share of manufacturing delays.
Reuters reported no reliable estimate of the secondary market’s share of the global turbine market, which is worth approximately $25 billion annually.
Prices reflect a complicated market rather than a simple discount for older equipment. Reuters cited indicative figures from market participants and researchers of $1.5 million to $1.7 million per megawatt for well-maintained used turbines, around $1.3 million for refurbished units, and $2 million to $2.5 million for new turbines. Actual terms and suitability vary. In one separate transaction, two 14.5-megawatt Solar Turbines units sold for $12.5 million combined.
Are you looking for developers?
Availability is only the first question. Buyers must consider maintenance histories, remaining service life, replacement parts, and long-term servicing arrangements. Converted jet engines also require evaluation for the demands of continuous power generation. An aeroderivative turbine is not automatically interchangeable with every conventional gas turbine.
When larger units cannot be sourced, developers may assemble plants from numerous smaller machines, sometimes more than 100 at one location, potentially combining different manufacturers and models. That approach changes the operational workload: more assets to inspect, maintenance schedules to coordinate, and equipment interfaces to manage.
Even securing equipment does not remove obstacles such as air permits and grid access. Nor is every alternative generation technology suitable for every project.
Gas-engine systems are attracting interest. Wärtsilä’s energy business president said the company had booked approximately three gigawatts of orders from US data centers since spring 2025. That is evidence of demand for another option, not proof that gas engines can universally replace turbines.
The broader point is that sourcing power for AI facilities has become a multi-part engineering problem. Generation technology, site requirements, electrical connections, and equipment support must be planned together, often amid uncertain delivery schedules.
Are you looking for developers?
Operating a facility with several generation technologies or many individual units creates information and coordination challenges. Teams may need to bring maintenance records, equipment status, alarms, and energy data into a usable operational view.
This is a general implication of more complex infrastructure, not a description of systems Crusoe has announced. Backend services, APIs, data pipelines, and monitoring interfaces can help connect information from different platforms. Alerting and maintenance workflows can support timely investigation, while testing and access controls help keep software changes manageable.
Such systems still require experienced operators and appropriate engineering oversight. Software cannot manufacture missing turbines, provide grid capacity, resolve permitting delays, or guarantee uptime. Its contribution is to make increasingly distributed operational information easier to use and maintain.
For businesses developing these connected systems, Square Codex can complement internal engineering teams. The Costa Rica-based nearshore software development partner serves North American companies through custom software, backend and API integration, data engineering, cloud development, DevOps, monitoring interfaces, and QA automation. Its role is software engineering, not designing or certifying generation equipment.
Crusoe’s turbine order illustrates how far AI infrastructure dependencies extend beyond the data hall. Project timelines now depend on an intersection of manufacturing capacity, power access, equipment condition, and operational integration. Even when electricity can be secured, making a complicated power environment workable remains a systems problem.